CN101782947A - Multimedia copyright protection method - Google Patents

Abstract

The invention relates to a multimedia copyright protection method, which can be used for the copyright protection of multimedia products such as digital images, audios, videos, and the like, and relates to the technical fields of image processing technologies, digital watermarking technologies, and the like. The method comprises the following steps of: firstly, transforming an image by adopting a wavelet lifting algorithm; then smoothing the image by adopting a cellular automaton; and finally, carrying out signature verification on the smoothed image. Proved by simulation result, the method has high robustness and can resist attacks of cutting, rotation, noise, and the like, and is suitable for the multimedia copyright protection in the fields of medicine, military, and the like.

Description

Multimedia copyright protection method

Technical field

The present invention relates to a kind of guard method of multimedia copyright, can be used for the copyright protection of media products such as digital picture, audio frequency, video etc., relate to technical fields such as image processing techniques, digital watermark technology.

Background technology

Along with popularizing and development of computer of network, media products such as digital picture, audio frequency, video bring great convenience to people.But because digital multimedia duplicates easily, duplicate the legitimate rights and interests of having damaged the copyright owner with the illegal distribution original works, digital multimedia presses for copyright protection unauthorized.

In order effectively to protect copyright, Chen has proposed the ChenShi algorithm, and this algorithm application DWT technology is carried out conversion to image, have certain robustness, but its calculated amount is big, and working time is longer.People such as Chang improve the ChenShi algorithm on the basis of Chen, have improved arithmetic speed, but the robustness of this algorithm is not good enough.Also the someone proposes to set up the copyright control center at present, but the copyright control center method that proposes at present still problem such as unmatchful authenticating safety analyse in depth, discuss.

Summary of the invention

This method has proposed a kind of multimedia copyright protection method of high robust on the basis of the above.This method is at first used the Lifting Wavelet algorithm image is carried out conversion, uses cellular automaton then image is carried out smoothing processing, at last the image after the smoothing processing is carried out signature authentication.

This method is based on the digital image copyright protection method of lifting wavelet transform, chaology, cellular automaton theory under authentication center control.

The authentication model that this method proposes mainly comprises signature and authenticates two steps.Signature step mainly is to handle multimedia image information, generates signing messages together with the logo image then.Mainly contain three roles in the pattern: signature embedding side (watermark embedding side), signature authentication side (watermark authentication side), authentication center.Signature embeddings side is the personnel that generate signature, and signature authentication side is any personnel that image is authenticated of wanting, authentication center is responsible for the relevant information that image is protected in the generation of signature embedding side.

Watermark embed process comprises as shown in Figure 1: adjust the multi-media image size, and the Lifting Wavelet image transformation, smoothing processing, the Logo image scrambling generates authentication image, generates steps such as authentication stamp.

At the picture that needs copyright control, at first carry out the lifting wavelet transform Wavelet Component that gets a promotion, this Lifting Wavelet component application cellular automaton is carried out smoothing processing obtain image O, this moment, image O was based upon on the basis of cellular automaton and lifting wavelet transform.This programme adopts Lifting Wavelet that original image is changed.Overcome the weakness of Fourier transform fixed resolution based on the multiresolution digital watermarking algorithm of first generation wavelet transformation, but both analytic signal general picture, but the details of analytic signal again, the good spatial frequency resolution characteristic is arranged, but its algorithm computation complexity, big to the internal memory space requirement, and the picture size requirement to some extent to doing conversion.Simultaneously, the coefficient that produces after the traditional wavelet is a floating number, because the restriction of computing machine limited wordlength, accurately reconstituting initial image.Lifting Wavelet is that the lifting by first generation wavelet transformation realizes that it is based on the wavelet construction method in spatial domain.The advantage of the existing traditional small echo of Lifting Wavelet has overcome its limitation again.Its fast operation, the complete original position of permission are calculated, and its inverse transformation is the inverse operation of direct transform.

At watermarking images, carry out the chaos scramble by chaotic key K and obtain image W, lower through the watermarking images W correlativity behind the chaos scramble, the telescopiny of whole watermark has been improved security.

Above-mentioned image O that obtains after treatment and image W are carried out XOR.Obtain authentication image V this moment.At this moment, the signature embedding can be to send an authentication request to authentication center.Send authentication image V simultaneously, image size N, threshold values T, the watermark side of embedding ID, chaos scramble key K.After request is received by authentication center, generate authentication according to formula and stab, extract watermark during user authentication process.When watermark authentication side need authenticate image, at first from authentication image, obtain V, N, T, ID, K} obtains authentication from authentication center and stabs h _TA, timestamp t, one-way Hash function H _TA(.), the multi-media image to the need authentication authenticates then.Verification process as shown in Figure 2.Verification process comprises following five steps: the authentication that authentication is stabbed, adjust the image size, and the Lifting Wavelet image transformation, smoothing processing is calculated the Logo image

Because image has passed through lifting wavelet transform, cellular automaton processing, chaos processing, therefore, this image can be resisted attacks such as cutting, rotation, compression, has higher robustness.Adopted the control of authentication center, guaranteed that this method compares additive method and have higher feasibility, security, reliability.

Simulation result shows that this algorithm has high robust, can resist attacks such as cutting, rotation, noise, is applicable to the copyright protection in fields such as medical science, military affairs.

Description of drawings

Fig. 1: signature step synoptic diagram;

Fig. 2: verification process synoptic diagram.

Embodiment

The authentication model that this method proposes mainly comprises signature and authenticates two steps.Signature step mainly is to handle multimedia image information, generates signing messages together with the logo image then.Mainly contain three roles in the pattern: signature embedding side (watermark embedding side), signature authentication side (watermark authentication side), authentication center.Signature embeddings side is the personnel that generate signature, and signature authentication side is any personnel that image is authenticated of wanting, authentication center is responsible for the relevant information that image is protected in the generation of signature embedding side.The specific implementation process is as follows:

One, signature process

Signature process is divided into 5 steps, comprising: adjust the multi-media image size, and lifting wavelet transform, smoothing processing, the Logo image scrambling generates authentication image, generates authentication and stabs, and implementation procedure is as shown in Figure 1.

1, adjusts the multi-media image size

The size of supposing multi-media image O is N * N pixel, at first multi-media image O is divided into mutually disjoint 8 * 8 block of pixels o _iSet, i.e. O={o ₁, o ₂..., o _{(N, 8) * (N/8)}.Authenticating party is responsible for calculating each piece o _iMean value m _i, i=1 wherein, 2 ..., (N/8) * (N/8).Obtain the downscaled images of the multi-media image O of (N/8) * (N/8) pixel size this moment.

2, lifting wavelet transform

The image that obtains at above-mentioned steps 1 carries out lifting wavelet transform, obtains image OL

The forward lifting process of Lifting Wavelet can be decomposed into three processes: division, prediction and correction, and the specific implementation process is:

(1) division.The original image sequence sets is decomposed into two littler subclass, λ _-1And γ _-1, be the odd and even number collection.

(2) prediction.λ _-1And γ _-1Represent odd and even number respectively, have certain correlativity, therefore can pass through λ _-1To γ _-1Predict, as shown in Equation 1:

γ _-1＝P(λ _-1)

So just can use λ _-1Represent sequence sets λ ₀Because λ _-1γ can not calculate to a nicety _-1, i.e. P (λ _-1) just approximate and γ _-1Equate, therefore, can use γ _-1And the difference between the predicted value is represented γ _-1, promptly formula 2:

γ _-1＝γ _-1-P(λ _-1)。

At this moment, can use less sequence λ _-1And γ _-1Represent original series λ ₀Above-mentioned steps is carried out iteration, promptly at first with λ _-1Decompose, obtain λ _-2And γ _-2, calculate new γ with formula 2 then _-2So repeatedly, constantly iteration can replace with { λ with original series _-n, γ _-n, Λ, γ _-1, this sequence is more compacter than original series.

(3) revise.In order to improve the reliability of said process, need revise data.The thinking of revising is to obtain one by algorithm Q to be more suitable in the λ of division _-1, formula 3:

Q(λ _-1)＝Q(λ ₀)

Consider the feasibility of algorithm, use γ herein _-1Upgrade λ _-1, promptly formula 4:

λ _-1＝λ _-1+U(γ _-1)

The above-mentioned algorithm of iteration can obtain formula 5,

$\mathrm{fori}=-1\mathrm{to}-n\left\{\begin{array}{c}\{{\mathrm{\λ}}_i,{\mathrm{\γ}}_i\}=\mathrm{Split}\left({\mathrm{\λ}}_{i+1}\right)\\ {\mathrm{\γ}}_i-=P\left({\mathrm{\λ}}_i\right)\\ {\mathrm{\λ}}_i+=U\left({\mathrm{\γ}}_i\right)\end{array}\right.$

3, smoothing processing

Image OL is carried out cellular automaton handle, to each pixel m ' among the image OL _iCarry out value again.Neighbours' model adopts the Moore model herein, and the local evolution rule is f1, as shown in Equation 7:

$s_i^{t+1}=\left\{\begin{array}{c}1,\mathrm{sum}\left(s_N^t\right)\&GreaterEqual;5\\ 0,\mathrm{sum}\left(s_N^t\right)<5\end{array}\right.$

According to above-mentioned formula, obtain new m ' _iThe value composition diagram is as E1.

4, Logo image scrambling

The size of supposing Logo image L is (N/8) * (N/8), application of chaos sequence

L1 carries out scramble to the logo image, obtains new Logo image L1.X value 0.63, value are 2.

5, generate authentication image

Image E1 and image L1 carry out xor operation, obtain new image V, as shown in Equation 8:

$V=E1\&CirclePlus;L$

6, generating authentication stabs

At first signature embeds direction authentication center and sends an authentication request.Send authentication image V simultaneously, image size N, threshold values T, the watermark side of embedding ID, chaos scramble key K.After request is received by authentication center, generate authentication according to formula 9 and stab.

h _TA＝H _TA(V||N||T||ID||K||t)

H wherein _TA(.) is one-way Hash function, and t is a timestamp, || represent connector.

After finishing above-mentioned steps, authentication center announces authentication stamp h _TA, timestamp t, one-way Hash function H _TA(.).

Two, verification process

When watermark authentication side need authenticate multi-media image, at first from the authentication multi-media image, obtain V, N, T, ID, K} obtains authentication from authentication center and stabs h _TA, timestamp t, one-way Hash function H _TA(.) authenticates original image then.Verification process as shown in Figure 2.Verification process comprises following five steps: the authentication that authentication is stabbed, adjust the image size, and generate the Lifting Wavelet image, smoothing processing is calculated the Logo image.

1, the authentication of authentication stamp

At first, authenticating party obtains authentication from authentication center and stabs h _TA, timestamp t, one-way Hash function H _TA(.).Then, authenticating party use obtain V, N, T, ID, K} and t carry out computing to formula (10).

$h_{\mathrm{TA}}^{*}=H_{\mathrm{TA}}\left(V\right|\left|N\right|\left|T\right|\left|\mathrm{ID}\right|\left|K\right|\left|t\right)$

Authenticating party is right

And h _TACompare, if they are equal, descriptive information V, and N, T, ID, K} are effective informations, otherwise information is invalid, need not authentication.

2, adjusting need the authentication image size

Be applied in the same technological adjustment image size of signature stage step 1, detailed process is: suppose that needing the size of authentication image is N * N pixel, at first will need authentication image to be divided into the set of mutually disjoint 8 * 8 block of pixels, that is, obtain the downscaled images of the multi-media image of (N/8) * (N/8) pixel size this moment.

3, generate the Lifting Wavelet image

The image that above-mentioned steps 2 is obtained carries out lifting wavelet transform, obtains image E '.

The forward lifting process of Lifting Wavelet can be decomposed into three processes: division, prediction and correction, and the specific implementation process is:

(1) division.The original image sequence sets is decomposed into two littler subclass, λ _-1And γ _-1, be the odd and even number collection.

(2) prediction.λ _-1And γ _-1Represent odd and even number respectively, have certain correlativity, therefore can pass through λ _-1To γ _-1Predict, as shown in Equation 1:

γ _-1＝P(λ _-1)

So just can use λ _-1Represent sequence sets λ ₀Because λ _-1γ can not calculate to a nicety _-1, i.e. P (λ _-1) just approximate and γ _-1Equate, therefore, can use γ _-1And the difference between the predicted value is represented γ _-1, promptly formula 2:

γ _-1＝γ _-1-P(λ _-1)。

At this moment, can use less sequence λ _-1And γ _-1Represent original series λ ₀Above-mentioned steps is carried out iteration, promptly at first with λ _-1Decompose, obtain λ _-2And γ _-2, calculate new γ with formula 2 then _-2So repeatedly, constantly iteration can replace with { λ with original series _-n, γ _-n, Λ, γ _-1, this sequence is more compacter than original series.

(3) revise.In order to improve the reliability of said process, need revise data.The thinking of revising is to obtain one by algorithm Q to be more suitable in the λ of division _-1, formula 3:

Q(λ _-1)＝Q(λ ₀)

Consider the feasibility of algorithm, use γ herein _-1Upgrade λ _-1, promptly formula 4:

λ _-1＝λ _-1+U(γ _-1)

The above-mentioned algorithm of iteration can obtain formula 5,

$\mathrm{fori}=-1\mathrm{to}-n\left\{\begin{array}{c}\{{\mathrm{\&lambda;}}_1,{\mathrm{\&gamma;}}_i\}=\mathrm{Split}\left({\mathrm{\&lambda;}}_{i+1}\right)\\ {\mathrm{\&gamma;}}_i-=P\left({\mathrm{\&lambda;}}_i\right)\\ {\mathrm{\&lambda;}}_i+=U\left({\mathrm{\&gamma;}}_i\right)\end{array}\right.$

4, smoothing processing

The same antialiasing of application signature stage step 3 generates smoothed image E1, and detailed process is:

Image E ' is carried out cellular automaton handle, each pixel among the image E ' is carried out value again.Neighbours' model adopts the Moore model herein, and the local evolution rule is f1, as shown in Equation 7:

$s_i^{t+1}=\left\{\begin{array}{c}1,\mathrm{sum}\left(s_N^t\right)\&GreaterEqual;5\\ 0,\mathrm{sum}\left(s_N^t\right)<5\end{array}\right.$

According to above-mentioned formula, obtain the E1 of new pixel value composition diagram picture.

5, calculate the Logo image

At first image E1 and authentication image V are carried out XOR, obtain image L '.Application of chaos system and key K then to the image L ' unrest that is inverted, obtain a Logo image L ^*At last, authenticating party can be to the image L that obtains ^*Recognize, thus the copyright of affirmation image.

Claims (4)

1. a multimedia copyright protection method is characterized in that, comprises signature and authenticates two steps, is specially:

A, signature step:

1), be that the media product image of N * N carries out the pixel adjustment with pixel size, obtain the downscaled images of the media product image of (N/8) * (N/8) pixel size;

2), at above-mentioned steps 1) image that obtains carries out lifting wavelet transform and obtains image OL;

3), image OL is carried out the cellular automaton processing, to each pixel m ' among the image E _iCarry out value again, obtain new m ' _iThe value composition diagram is as E1;

4), the application of chaos sequence carries out scramble to Logo image L and handles, and obtains new Logo image L1;

5), image E1 and image L1 carry out xor operation, obtains new image V;

6), signing embeds authentication request of direction authentication center transmission, sends authentication image V simultaneously, image size N, threshold values T, the watermark side of embedding ID, chaos scramble key K.After request is received by authentication center, generate authentication and stab h _TA

7), authentication center announces authentication stamp h _TA, timestamp t, one-way Hash function H _TA();

B, verification process:

1), authenticating party obtains authentication from authentication center and stabs h _TA, timestamp t, one-way Hash function H _TA(), use obtain V, N, T, ID, K} and t are to formula

Carry out computing, to h _TA* and h _TACompare, if they are equal, descriptive information V, and N, T, ID, K} are effective informations, otherwise information is invalid, need not authentication;

2), adjust to need the authentication image size, suppose that the size of need authentication image is N * N pixel, will need authentication image to be divided into the set of mutually disjoint 8 * 8 block of pixels, obtain the downscaled images of the multi-media image of (N/8) * (N/8) pixel size;

3), to above-mentioned steps 2) image that obtains carries out lifting wavelet transform, obtains image E ';

4), image E ' is carried out the cellular automaton processing and obtain smoothed image E1;

5), image E1 and authentication image V are carried out XOR, obtain image L '.Application of chaos system and key K then to the image L ' unrest that is inverted, obtain a Logo image L ^*, last, authenticating party can be to the image L that obtains ^*Recognize, thus the copyright of affirmation image.

2. method according to claim 1 is characterized in that comprising in the method three roles: signature embedding side, signature authentication side and authentication center.

3. method according to claim 1 is characterized in that lifting wavelet transform comprises three steps: division, prediction, correction.

4. method according to claim 1 is characterized in that the cellular automaton processing is preferably the Moore model.

Images